ABSTRACT The microbial environment in the human vagina is extremely dynamic. Bacterial communities can shift dramatically from lactobacillus predominance, to a diverse, polymicrobial state, which is a fundamental characteristic of bacterial vaginosis (BV). BV is a clinically important condition associated with vaginal discharge as well as heightened risk of preterm birth, pelvic inflammatory disease, post-surgical infections, HIV and other sexually transmitted infections. Unfortunately, little is known about the factors that prompt transitions between simple and complex microbial communities and in particular, why BV commonly recurs following antibiotic treatment. The purpose of this proposal is to use complementary tools from multiple disciplines to describe the influence of environmental factors and interspecies interactions on vaginal bacterial dynamics. Our focus will be on characterizing factors underlying BV recurrence, maintenance and resolution. We hypothesize that threshold concentrations of certain species and nutrients are necessary for recurrent BV, while different conditions, in particular establishment of biofilms, are required for maintenance of this anaerobic, polymicrobial state. In Aim 1 we will define the natural history of bacterial kinetics by collecting vaginal swabs in women 3 times per day for 60 days, to measure abundance of specific bacterial species with qPCR. We will also collect daily behavioral diaries, and sample for pH and nutritive biochemicals such as hemoglobin and glucose which are known to vary with hormonal cycling. In addition, we will perform broad-range 16S rRNA gene PCR with pyrosequencing to describe global changes in microbial community profile diversity. We will therefore be able to identify the key microbial and nutrient conditions that predict abrupt transitions in vagina microbiota. In Aim 2 we will measure bacterial growth, competition, and biofilm behavior under selective metabolic conditions using a realistic microfluidic in vitro cultivation system. We will manipulate key bacterial metabolites such as amino acids, host derived nutrients such as glucose and iron, environmental conditions such as pH, and other biophysical features of the vagina such as shear forces and biofilm formation, to assess the effect of these variable on microbial growth rates, as well as microbial competition among multiple species for scare nutrients. In Aim 3 we will use findings from the laboratory experiments (Aim 2) to populate mathematical models with competing assumptions that describe bacterial community dynamics. Models will be tested for their ability to reproduce key dynamic features from human longitudinal data in Aim 1. The optimal model will generate mechanistic explanations for observations from Aims 1 and 2, and will identify conditions necessary for development, maintenance and eradication of BV. A greater understanding of the necessary conditions for abrupt microbial shifts will offer pathways for improving reproductive health.